Network Function Virtualization – Interview with Benedetto Pietrabella, Chief Architect at Xura
The cultural phenomenon that surrounded the Short Message Service (SMS) and texting boom pretty much secured revenue for Communication Service Providers (CSPs) for more than a decade. While the texting market continues to be a major revenue contributor, CSPs increasingly find themselves facing new revenue retention challenges from competition such as the “freemium” cloud-based messaging offerings of Over-The-Top providers (OTTs) like WhatsApp and Facebook Messenger. In addition, as hardware reaches end of life, CSPs are increasingly being forced to evolve or replace infrastructure, urging them to make decisions on how best to support existing services, including new IP-based next generation messaging. As operators are having to upgrade their systems, an NFV environment provides the opportunity to break the tie between software and hardware and move into a virtualized environment that can run in data centers with heterogeneous infrastructure. For this reason, many are looking to implement Network Function Virtualization (NFV) for service delivery including messaging while turning themselves into cloud service providers.
In the first quarter of 2016, Xura and Heavy Reading carried out a detailed survey to assess the progress, market timing, business and technical drivers that CSPs are facing as they look to virtualize their network, with a particular focus on messaging services. The survey contained a total of 19 questions and was answered by over 70 qualified respondents in the CSP space.
Are CSPs interested in virtualizing their networks? What are the benefits of implementing NFV from a CSPs perspective?
BP: As CSPs are faced with the reality of modernizing their networks to meet rising market demand for new IP-based services, and compete in a world of OTT services, many have taken bold steps to virtualize their networks, with momentum behind NFV growing rapidly since October 2013 – when the first set of specifications were defined. Fast-forward to 2016, and 70 percent of CSPs plan to virtualize some, or all, of their platforms at some point in the next four years. According to our survey, U.S. respondents, in particular, have aggressive timelines for NFV implementation, aiming for one to two year timeframes.
The benefits of NFV implementation are undeniable. The near-term payback of lower capital and operational expenditure by opting for virtualization, including data center costs, form the business case for funding the investment. Longer term, the impact of increased flexibility will radically transform the operational business of running a telco. NFV will provide operators with a much greater degree of network elasticity and the agility to adapt to market or consumer behavioral changes simply and quickly. Existing competition and new entrants will always continue to disrupt the future landscape so CSPs need to be ready to overcome challenges and meet expectations from shifting trends.
The operational efficiencies of NFV functionality such as multi-tenancy, lifecycle management and service chaining have the potential to strategically boost CSPs’ competitive position relative to OTT players and help CSPs introduce new, innovative services at a faster speed, lower costs and with more attractive pricing to consumer and enterprise markets alike.
What is the leading driver of NFV from a technical vs. business perspective?
BP: The overwhelming driver behind NFV from a technical perspective is operational efficiency, with respondents ranking it top of the list as a factor relating to SMS, MMS, IP Messaging, RCS, WebRTC, Voicemail systems, spam/fraud messaging control and control plane security services (32%-55% range). Network scalability and elasticity followed (17% – 34%), with application consolidations ranking third (13% – 19%) and enhanced security overall, falling into the last (3% – 13%) range, with the exception of Spam/fraud messaging control (where security was, not surprisingly, the dominant driver).
From a business perspective, presented with the same suite of messaging services, respondents found the leading driver behind implementing NFV to be service agility and flexibility (22% – 28% range). However, it’s clear that CSPs are looking to reap the lower OPEX and capex cost benefits of the cloud as well. As a result, other business drivers, such as hardware-related OPEX reduction (19% -29% range) and capex reduction (16% – 30% range), also scored highly in the survey.
Overall, these findings reinforce the multifaceted value proposition that CSPs place on the value of the virtualization of messaging services. Put another way, CSPs understand the value of service agility, but they are now pragmatically looking for other business benefits to justify commercialization.
What are some of the top priorities for short-term virtualization?
BP: It’s clear that NFV implementation is a top priority for most CSPs, and that when deployment decisions are being made, there are short-term priorities that should be considered. According to our results, in the short term CSPs are focused on SMS (27%), spam/fraud messaging control (22%), followed by IP messaging/Unified Communications (21%).
Virtualizing SMS with elastic scaling has huge implications for occurrences such as “signaling storms” that are created by severe spikes in SMS traffic. To deal with the great demands that are seen during holiday periods, sporting events or in times of disaster, operators are currently bringing in extra expensive stacks to facilitate the function. Wouldn’t it be better to have the agility of an auto-ready NFV SMS solution that can flexibly scale in real-time?
While focus on SMS is a logical starting point in the virtualization roadmap, the very close scoring of the other short-term priorities means that CSPs see a tangible value proposition for virtualizing all messaging platforms. Xura is currently working with global operators on deploying virtualized and NFV based solutions across all messaging and voice mail solutions.
What are the challenges and impediments to widespread NFV implementation?
BP: While the drivers for virtualizing messaging solutions are powerful, there are several challenges and obstacles to widespread implementation. On the technical side, CSPs consider product interworking (17% – 36% range), orchestration (14% -27% range) and migration complexity (13% – 17% range) as the most formidable challenges across all messaging services. Respondents scored security relatively low as a top challenge, with only a 21 percent spike here related to spam/fraud messaging.
However, the question of product interworking will not be an issue in the near term due to well-defined standard interfaces for many existing services across the mobile ecosystem. For example, SMS works in a granular way, as well as MMS. This means the application works in the same way whether it’s virtual or not; so the move to NFV and product interworking should be seamless. In the longer term, as service chaining becomes more prevalent we can expect that well defined interfaces will be developed across Virtual Network Functions that can be integrated across vendor products to create new agile services.
A few other items come to mind when we deal with orchestration products, deploying services or Virtual Network Functions (VNF’s) in the cloud rather than on bare metal. For one, orchestration is still relatively new for telecommunications deployment scenarios, so it is understandable CSPs are nervous. For another, there is concern that CSPs will deploy multiple vendor-specific VNF orchestrators (rather than implementing a single orchestrator), which may inject both additional cost and complexity into any virtualization migration. Interestingly, U.S respondents (23% – 47% range) rated orchestration a significantly higher barrier to overcome than rest of world respondents (RoW) (6% – 11% range).
On the business side, respondents anticipated cultural challenges (35% – 51%) and business case definition (27% – 46%) to have the most impact on migration. Although we see that operators expect to reduce CAPEX and OPEX from migration, business case detailed definition remains complex as it is a combination of equipment and resource savings as well as operational process improvement. Reduction in these expenses will come once more of the services have been migrated, but it’s hard to justify savings as a business case when only one messaging service has moved over. It’s a bit like changing the tires on your car as it’s driving along. When looking at cultural challenges, one has to consider NFV’s disruptive impact on an organization’s culture, service implementation and delivery, personnel training as well as, who will manage the new services on a daily basis.
How should CSPs strategically employ NFV technology?
BP: While there were no specific questions asked in our survey around the most strategic way to implement NFV, we at Xura understand that there are several different ways CSPs can successfully and strategically implement NFV technology. CSPs should consider their existing legacy system capabilities, reasonable timelines, available resources and long and short term priorities before delving into a complete NFV overhaul. Every CSP will have their own individual timing and approach to the project – so it’s important to keep this in mind when building the overarching NFV strategy.
Some operators may decide to start with smaller deployments and grow the environment instead of migrating existing services. New services such as VoLTE or IP-SM for Wi-Fi SMS could be tested in the environment without the pressure of millions of subscribers. Regardless of what and when; one thing is clear, operators understand the need to modernize their networks, and optimizing their hardware footprint is an essential step towards running a much more efficient network.
In addition, it’s important to look at the support and deployment models available. Operators may choose to either have a managed service within their data center or directly manage the NFV environment themselves within their data centers. Operators want a new infrastructure and still want to control the platform while providing Telco grade availability and not deploy it to the public cloud. Xura’s solutions support in-network as well as cloud based deployments. This makes us an appealing vendor for delivering NFV based SMS, MMS, RCS and Voicemail solutions that are future proof and can grow feature functionality within the NFV ongoing roadmap.
What are the main design considerations CSPs should be aware of when implementing NFV? How should they implement technologies like OpenStack with NFV?
BP: For network operators embarking on an NFV project, the two approaches that seem to dominate the conversation are OpenStack and VMware. According to survey respondents, the preferred approach to implementing virtualization technologies is to implement OpenStack supplied and integrated by a vendor, like Nokia or Cisco (49%). Additionally, respondents noted integrating OpenStack with contractors (44%) and from an integration, like IBM or HPE (41%), was preferred. VMware also had considerable support with 41% of respondents. It’s clear that OpenStack has entered the market and is becoming the defacto standard. We see a trend toward stable and supported Open Source based solutions as providing the most flexibility to operators. It is clear that products that are part of the MANO stack such as OpenStack must be easy to integrate and have open and standard interfaces.
What are CSPs considering when choosing NFV based product vendors?
BP: For all CSPs, choosing an NFV vendor is a big decision with many moving parts. According to our survey, security and scalability (both 76%) came in first as the top considerations when choosing a vendor. Virtualization product roadmap (59%), RFP compliance (49%), and solution openness (47%) were other things respondents found compelling. Considering this, it’s clear that CSPs are looking for vendors with mature products that are able to deliver solutions that align with the functionality required of NFV.
We at Xura consider this last point especially important, because it shows that even though many CSPs are willing to adopt vendor-specific orchestrators, they still consider the ability to support third-party orchestrators as critical to maintain operational efficiencies.
Perhaps the most compelling result is the high ranking of security. CSPs understand the importance of securing cloud deployments and rank it highly in several specific use cases throughout our study, but overall it seems to be smoldering beneath other priorities. All in all, this shows that CSPs plan to heavily rely on their vendors to help them implement the measures necessary to secure their virtualized networks.
What is a realistic timeframe for CSPs to utilize NFV?
BP: The move to NFV is a complete evolution of incumbent bare-metal networks and this reimagining of deploying and managing network services will not happen overnight, with typical projects ranging anywhere from one to several years. This means that there will likely be a long period of time where operators must achieve interworking between their legacy and virtualized network functions so they can continue serving customers without interruption. This will surely increase the complexity of operations resulting as a result of a hybrid network topology that requires sophisticated abilities to bridge physical and virtualized functions and perform service routing and balancing.
Will consumers see any benefit from increased NFV?
BP: Although subscribers may not immediately physically see anything different with their mobile devices and networks, increasing business agility will create a path for new commercial opportunities. This will result in cost savings passed onto subscribers from mobile operators. More likely we will start to see new “innovative” services being deployed in this new and faster agile product launch. These solutions will step up to provide consumers with new services that will represent the next frontier of ultimate experience in consumer to consumer or consumer to business mobile applications.